Particle-based Model of full-size ITER-relevant

Negative Ion Source

Francesco TaccognaPierpaolo MinelliNicola Ippolito

P.Las.M.I lab@CNR-NanotecINFNBari

e-mail: francesco.taccogna@cnr.it 16th ICIS, 25th August 2015

Outline

o Status of present models: Weakness vs Strength

o Full-size Source Model with detailed extraction: MINUS

o Preliminary results

o Conclusions

3

NIS Models

EXTRACTION ZOOMED MODEL FULL-SIZE SOURCE MODEL

Simplifying 2 different kind of approaches are present

y

z

Driver

4

NIS Models

EXTRACTION ZOOMED MODEL

+ Fine resolution (aperture contains tens of cells)+ Good statistical level (number of part. er cell)+ Realistic sheath representation (real vacuum permittivity)+ 3D model+ Detailed electron deflection field+ Surface-produced negative ions+ Optimization of aperture size/shape+ Meniscus shape and beam optics

- One aperture (periodic conditions)- Strong influence of free parameters (initial conditions, re-injection system)- Anisotropic fluxes and possible non-Maxwellian behavior- Magnetic filter effects starts much farther from PG

NIO1 extraction type: see Poster TuePE30

5

NIS Models

FULL-SIZE SOURCE MODEL

+ Model of the plasma expansion and electron transport across the magnetic filter field+ Dis-homogeneity along PG+ Self-consistent plasma condition at the entrance of the extraction region+ Positive ion flux at PG (ion conversion)+ BP and PG bias effect+ Plasma-gas coupling+ 3D model

- Scaled model (fake vacuum permittivity)- Cell size larger than the single aperture - Driver not self-consistently included

z

Driver

6

NIS Models

EXTRACTION ZOOMED MODEL

+ Fine resolution (aperture contains tens of cells)+ Good statistical level (number of part. er cell)+ Realistic sheath representation (real vacuum permittivity)+ 3D model+ Detailed electron deflection field+ Surface-produced negative ions+ Optimization of aperture size/shape+ Meniscus shape and beam optics

- One aperture (periodic conditions)- Strong influence of free parameters (initial conditions, re-injection system)- Anisotropic fluxes and possible non-Maxwellian behavior- Magnetic filter effects starts much farther from PG

FULL-SIZE SOURCE MODEL

+ Model of the plasma expansion and electron transport across the magnetic filter field+ Dis-homogeneity along PG+ Self-consistent plasma condition at the entrance of the extraction region+ Positive ion flux at PG (ion conversion)+ BP and PG bias effect+ Plasma-gas coupling+ 3D model

- Scaled model (fake vacuum permittivity)- Cell size larger than the single aperture - Driver not self-consistently included

Simulation domainExpansion + Extraction + 1st acceleration step multiaperture grid (10 apertures):BP, PG and EG included.

2.5D PIC-MCC full source Model: minus.f

24 cm

58 c

m

y

zBias plate

PG

EG

Driver

Assumptions-Limitations• Driver/InjectionThe driver is not yet simulated:prescribed ambipolar neutral full-maxwellian flux of plasma particleinjected in a thin area located at the driver exit plane

2.5D PIC-MCC full source Model: minus.f

24 cm

58 c

m

y

zBias plate

PG

EG

Driver

ne=6x1017 @ Te=12 eV

nH+=2.4x1017 @ TH+=1 eV nH2+=3.6x1017 @ TH2+=1 eV

Assumptions-Limitations• Driver/Injection• Geometry2.5D cartesian geometry - 2D(y,z) electrostatic [1]

Particle tracked in x (magnetic filter field direction) but quantity considered uniform (Ex=0)Particle-Wall interaction are considered assuming a thin sheath approximation;A secondary electron emission coefficient ϒ=0.2 has been assumed

2.5D PIC-MCC full source Model: minus.f

24 cm

58 c

m

y

zBias plate

PG

EG

Driver

[1] Poisson equation solver: http://www.mcs.anl.gov/petsc/petsc-as/

Assumptions-Limitations• Driver/Injection• GeometryScaled model: ε’0=25ε0 Δt’=5Δt;

Δz’=5ΔzIt allows keeping the detailed mesh of one-aperture modelThe aperture (D=10 mm, flat) contains 25 cells

2.5D PIC-MCC full source Model: minus.f

24 cm

58 c

m

y

zBias plate

PG

EG

Driver

Assumptions-Limitations• Driver/Injection• Geometry• Input data- Filter field: bell shaped z-profile with zmax=3 cm from PG and Bx,max=7 mT- Electron deflection field: prescribed 2D(y,z) map with alternating y-direction for adiacent apertures- Fixed H and H2 density with vibrational Boltzmann distribution

2.5D PIC-MCC full source Model: minus.f

24 cm

58 c

m

y

zBias plate

PG

EG

Driver

Assumptions-Limitations• Driver/Injection• Geometry• Input data- Most relevant collisions included- Self-consistent production of volume H- and surface by ion conversion [1]

- Fixed current density J=660 A/m2 of surface-produced H- by neutral conversion; uniformly launched along y at PG

2.5D PIC-MCC full source Model: minus.f

[1] M. Seidl, H.L. Cui, J.D. Isenberg, H.J. Know, B.S. Lee, S.T. Melnychuk, J. Appl. Phys. 79, 2896 (1996).

JH-,0=660 A/m2

BP

PG

EG

2.5D PIC-MCC full source Model: minus.f

24 cm

58 c

m

y

zBias plate

PG

EG

Driver

Simulation parameters-Performances• t = 2.5x10-11 s• x(=4x10-4 m) ~ D -> Ng=Ny x Nz=1450x589• Npart = 2x108 (w=2x108)• OpenMP/MPI hybrid paradigm• Ttot = 0.5 s in 7 days @ 4 Otta-Core Intel XeonE5/2640.v2 (2 GHz, 64 GB RAM)

Results: Temporal Evolution of Extracted Currents

- Steady state after probably 30 s; at the moment the simulation has reached 13 s;- The surface production is activated after 10.5 s and the extracted H- ions jumps from 40 A/m2 (volume contribution) to 200 A/m2;- H- ions from ion conversion on PG not yet produced;- Co-extracted electron current evolution has an intermittent behavior due to a turbulent transport across the filter typical of low pressure discharges (Magnetron Hall-effect thruster, etc.);

H- surface production starts

Results: Electric Potential

- Bottom-top y-asimmetry starting from the filter region;Hall and diamagnetic drifts contributes to create a space charge unbalanced along y and an electric field self-consistently develops to restore the plasma neutrality

- Meniscus penetrates too much because plasma is not yet arrived to screen EG field;

- Bulk value 10 V increasing;

Results: Electric Potential

- Bottom-top y-asimmetry starting from the filter region;Hall and diamagnetic drifts contributes to create a space charge unbalanced along y and an electric field self-consistently develops to restore the plasma neutrality

- Meniscus penetrates too much because plasma is not yet arrived to screen EG field;

- Bulk value 10 V increasing;

- Electric potential modulation along y at the filter region-> anomalous cross-field electron trasnport

Results: Electric Potential - Zoomed view of extraction region

- Plasma from the expansion region not yet able to screen the EG field

- Nevertheless plasma potential asymmetry is yet evident along the different apertures

- Meniscus shape is already forming

- The potential well attached to PG has a depth of 5 V

Electron Flux / Temperature

Extracted current aperture by aperture

Bias plate

PG

EG

Je (A/m2) JH- (A/m2)

0.5 178

0.6 263

0.7 111

0.6 146

0.8 156

1.4 153

1.3 145

1.4 140

9 255

244 194

First attempt to develop a 2.5D realistic PIC-MCC model of the full-size source with detailed extraction region

Keep the fine mesh used for the extraction region (cell size even smaller)

The work is going in 2 directions: - inclusion of the IC driver module - speed-up the execution of the code by programming optimization - application to NIO1

Preliminary results (not yet steady state reached) show: - the presence of y-modulation of electric potential responsible for the

electron cross-field anomalous transport in the filter field region - y-dishomogeneity along the PG inducing an electron extraction current - the electrostatic nature of the negative ion extraction is confirmed - about 30% of the surfaced-produced negative ions is extracted - the extracted negative ion current rise up 5 times when the surface

production is activated.

Conclusions